You can locate any trailer in the fleet to the metre and still not know what happened at the dock door. Warehouse yard video closes that gap — but it earns its keep on the daily dispute, not the rare theft.
Telematics ends at the gate. Head office can tell you where every vehicle is, how fast it went, how long the engine idled and when the door was opened — and then the trailer reverses onto a dock face six hundred kilometres away and the data goes quiet. What happened in the next ninety minutes exists only as a signature, a timestamp typed by somebody, and a phone call three days later when the numbers don't agree.
That gap is what warehouse yard video is for. And the reason most yard camera projects underdeliver is that they get bought as security systems, when the questions head office actually cannot answer are almost entirely operational.
It's worth listing them, because the list decides the whole design.
The Blind Spot Is Operational, Not Criminal
Theft at a yard is real and it is expensive. It is also rare, in the sense that any given site goes a long time between incidents. Meanwhile the following happen constantly: a load arrives short and nobody can say whether it left that way; a customer claims damage that the warehouse insists happened in transit; a carrier bills for detention the site says it never caused; a driver's arrival time and the gate log disagree by forty minutes; a trailer leaves with a seal nobody photographed.
Every one of those is a dispute, every dispute has a cost, and every one of them is decided by whoever can produce a credible account of what happened and when. Today that account is usually a person's memory and a handwritten line in a book.
So the honest business case for yard video is not that it will catch a thief. It is that it converts a category of recurring, unwinnable arguments into questions with answers. Buy it for the thing that happens every day, and the rare event is covered as a side effect. Buy it the other way around and you will build a system optimised for an event that hasn't happened yet.
Which sets a requirement most yard installations quietly fail.
Every One of Those Questions Is a Timestamp Question
Look at what actually settles a dock dispute. Not an image — an image with a time on it that the other party cannot dismiss. “The seal was intact at 14:07 and the trailer left at 14:12” only wins the argument if 14:07 means something.
This is where yard systems tend to be weakest, because a camera's clock is nobody's job. Left unsynchronised, a camera free-runs and its error accumulates from whenever it was last set, silently, with nothing reporting the drift. And conformance doesn't rescue you: the ONVIF Core Specification defines the interfaces for setting device time and NTP configuration but states no accuracy requirement for the device clock at all — the mechanics of which are covered in the multi-camera synchronisation piece.
The second failure is subtler and specific to reconstructing a sequence. A yard dispute usually spans cameras — gate, then yard, then dock face — and the RTP timestamps those streams carry cannot be compared with each other, by design. So if your recorder stamps frames when they arrive rather than when they were captured, you have baked every network delay in the yard into your evidence, permanently, and a forty-second disagreement between the gate camera and the dock camera is enough to lose an argument about whether a driver was made to wait.
Fix it cheaply: one time source on the local network, every camera pointed at it, capture timestamps carried through to storage rather than receipt times, and a scheduled check that alerts when any camera's reported time drifts. None of that is expensive. All of it is the difference between footage and evidence.
With the clocks honest, the next question is how much of this needs to be live at all.
Which Questions Actually Need Real Time
Real-time is the headline on most yard video pitches, and it deserves a harder look, because the majority of the value above is archival.
| The question head office asks | Needs live? | What it actually needs |
|---|---|---|
| Was the trailer sealed when it left? | No | Archive with a trustworthy timestamp |
| Did the driver arrive when the log says? | No | Gate footage the clock can be trusted on |
| Was that pallet damaged on our floor? | No | Dock-face coverage, retained past the claim window |
| Why did that truck wait two hours? | No | A reconstructable sequence across cameras |
| Is bay 4 free right now? | Yes | Low-latency view, low resolution is fine |
| Is the driver at the gate now? | Yes | Live view plus a way to talk to them |
| Is that manoeuvre unsafe, right now? | Yes | Live view someone is actually watching |
The split is roughly four to three in this list and heavily archival in practice, because the archival questions occur many times a day per site and the live ones occur occasionally. That is not an argument against live view — the live rows are genuine, and a yard controller who can see that bay 4 is empty stops making a driver wait for no reason. It is an argument against paying for a real-time architecture across every camera in the yard when a subset needs it.
The practical shape that follows: a low-resolution live view available on demand for the decision points, a full-quality continuous recording underneath everything, and the ability to pull either from head office without anyone driving to site.
Which is where the network reality of a yard asserts itself.
What “From Head Office” Actually Costs
A distribution network is a multi-site problem before it is a camera problem. Each site has its own uplink, usually sized for ERP traffic and not for video, and the instinct to stream every camera back to a central office is the design that fails first — connecting many sites to one operations centre isn't a bandwidth problem, it's a question of what gets relayed at the edge and what only moves when somebody asks for it.
Yards also punish infrastructure in ways an office building doesn't. Distances exceed comfortable copper runs, so you are into fibre or wireless links across open ground. Cameras sit outdoors through weather, at angles that face directly into low sun twice a day. Vehicles block sightlines that were clear at survey. And the site's own network is often flat, old, and administered by somebody whose main job is the WMS.
Nor will you be starting fresh. Nearly every yard already has cameras on a recorder that nobody outside the site can reach, and the useful move is usually to take a feed from what exists rather than replace it — which is a design decision in itself, covered in where to tap the stream without fighting the NVR.
All of which resolves into a handful of decisions worth making deliberately.
Designing Warehouse Yard Video That Answers Questions
- Cover decision points, not acreage. The gate, the seal, the dock face, the trailer parking rows. A camera watching empty tarmac generates storage; a camera watching the seal being applied settles claims. Site the cameras from the list of disputes you actually have.
- Give every camera the same clock, and check it. One time source on the local network, capture timestamps carried into storage, and an alert on drift. This is the cheapest thing in this article and the one that decides whether the footage is worth anything in an argument.
- Separate the live path from the archive path. Low-resolution live for the handful of now questions, full-quality continuous recording for everything else. Trying to serve both from one full-resolution stream is what makes yard video expensive.
- Retain past the claim cycle, not past the incident. Damage and detention disputes surface weeks after the event, so retention is set by your commercial terms rather than by a security policy — and the storage and governance side of that is its own problem, covered in recording at scale and the compliance trap.
- Give operations access, not just security. If the only people who can pull yard footage sit in a security function, the daily operational value never materialises. The yard controller, the claims team and the transport planner are the users who will justify the system.
- Add audio at the gate before you add cameras anywhere else. Most gate interactions are a question that needs answering — the ability to speak to a driver from a desk removes more delay per pound than another camera angle will.
Which leaves the question of what to build it on.
Build, Buy, or Deploy
- Build on open source. Media servers like go2rtc or MediaMTX at each site handle RTSP ingest from the existing cameras and deliver low-latency video to a browser, with recording alongside. You own the per-site cost, which matters across a network of depots. Right when you have engineering and the estate is large enough to justify it.
- Use a managed cloud relay. Quick to get a head-office view working with nothing to run per site — but per-stream pricing across every camera in every yard is the curve that ends these projects, and continuous recording to a third party's cloud gets expensive faster than anyone models.
- Deploy a commercial platform. A full stack — per-site ingest, low-latency live view, recording and retention — run on-premise or as a managed cloud deployment under a flat license. Samvyo is one such option: based on SFU architecture, delivering live yard and dock views to a browser at sub-second latency, with the media path, TURN and recording under your control, which matters when the archive is what settles commercial disputes. The transport question underneath it is getting RTSP cameras into a browser at all. Where it doesn't fit: a single site where somebody can walk to the dock and look.
Three routes, one test — can head office answer the question without phoning the site?
The Bottom Line
Fleet telematics stops at the gate, and the ninety minutes on the other side of it is where loads go short, damage gets disputed and detention gets billed. Yard video closes that gap, but only if it's bought for the argument that happens daily rather than the theft that happens rarely — and only if the timestamps survive scrutiny, because a dock dispute is settled by a credible time, not by a picture. Give every camera one clock, store capture times rather than arrival times, cover the decision points instead of the tarmac, retain past the claim cycle rather than the incident, and put the footage in the hands of operations rather than only security. Do that and most of the value arrives without anyone watching a screen — which is the opposite of how yard video usually gets sold.
What's Next?
The clock problem underneath every dispute in this article — why independent cameras drift and what actually fixes it — is the subject of the multi-camera synchronisation piece.
For the live half, and how much delay a response can absorb before it stops being real time, see Talk-Down at Ten Seconds Is Theater.
And for whether the alerts a site generates can realistically be watched by anyone, the operator-to-camera ratio does the arithmetic.
Frequently Asked Questions
What is warehouse yard video used for?
Mostly for settling operational disputes rather than catching theft: proving a trailer was sealed, establishing when a driver actually arrived, showing whether damage occurred on site, and reconstructing why a vehicle waited. Those happen daily, whereas theft at any given site is infrequent, so the archival value usually outweighs the live monitoring value.
Do I need real-time video at the loading dock?
For a subset of questions, yes — whether a bay is free, whether a driver is waiting at the gate, whether an unsafe manoeuvre is happening now. Most yard questions are answered from the archive afterwards. A sensible design gives you low-resolution live view on demand at the decision points, with continuous recording underneath everything.
Why do camera timestamps matter so much in a yard?
Because a dock dispute is won with a credible time, not just an image. Unsynchronised cameras free-run and accumulate error silently, and ONVIF conformance guarantees nothing here — the specification defines interfaces for setting device time but states no accuracy requirement for the clock. One shared time source and stored capture timestamps are what make footage usable as evidence.
Can I use the cameras already in the yard?
Usually yes, and it's normally the right move. Most yards already have cameras feeding a recorder that nobody offsite can reach, so the work is taking a feed from the existing estate and delivering it to a browser rather than replacing hardware. Where you tap that stream is a design decision with real consequences for load and latency.
How do I get yard video to head office without saturating the site link?
Don't stream everything centrally. Relay at the edge, keep full-quality recording on site, and move video to head office only when someone requests it or when a defined event fires. Site uplinks are sized for business systems, not continuous video, so central architecture is what decides whether the system scales across depots.
How long should yard footage be retained?
Long enough to cover your commercial claim window, which is typically longer than a security retention policy would suggest. Damage and detention disputes often surface weeks later, so retention should be set from your contractual terms — with the storage cost and data-governance implications planned rather than discovered.
What infrastructure does multi-site yard video need?
Per-site ingest from existing cameras, edge relay so uplinks aren't saturated, browser-deliverable live view for the decision points, and retention that survives the claims cycle. Samvyo is based on SFU architecture and keeps the media path, TURN and recording under your control, which matters when the archive is what settles commercial disputes.
